Page 1
HAL Id: hal-03186880https://hal.archives-ouvertes.fr/hal-03186880
Submitted on 7 Apr 2021
HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.
Conductance switching of azobenzene-basedself-assembled monolayers on cobalt probed by UHV
conductive-AFMLouis Thomas, Imane Arbouch, David Guerin, X. Wallart, Colin van Dyck,
Thierry Melin, Jérôme Cornil, Dominique Vuillaume, Stéphane Lenfant
To cite this version:Louis Thomas, Imane Arbouch, David Guerin, X. Wallart, Colin van Dyck, et al.. Conductanceswitching of azobenzene-based self-assembled monolayers on cobalt probed by UHV conductive-AFM.Nanoscale, Royal Society of Chemistry, 2021, 13 (14), pp.6977-6990. �10.1039/D1NR00106J�. �hal-03186880�
Page 2
Conductance Switching of Azobenzene-Based Self-Assembled
Monolayers on Cobalt Probed by UHV Conductive-AFM.
Louis Thomas,1,# Imane Arbouch,2,# David Guérin,1 Xavier Wallart,1 Colin van Dyck,2
Thierry Mélin,1 Jérôme Cornil,2,* Dominique Vuillaume1,* and Stéphane Lenfant1,*
1) Institute of Electronics Microelectronics and Nanotechnology (IEMN), CNRS, University of Lille,
Avenue Poincaré, Villeneuve d'Ascq, France.
2) Laboratory for Chemistry of Novel Materials, University of Mons,
Place du Parc 20, Mons, Belgium.
ABSTRACT
We report the formation of self-assembled monolayers of a molecular photoswitch (azobenzene-
bithiophene derivative, AzBT) on cobalt via a thiol covalent bond. We study the electrical properties
of the molecular junctions formed with the tip of a conductive atomic force microscope under ultra-
high vacuum. The statistical analysis of the current-voltage curves shows two distinct states of the
molecule conductance, suggesting the coexistence of both the trans and cis azobenzene isomers on
the surface. The cis isomer population (trans isomer) increases (decreases) upon UV light irradiation.
The situation is reversed under blue light irradiation. The experiments are confronted to first-
principle calculations performed on the molecular junctions with the Non-Equilibrium Green’s
�1
Page 3
Function formalism combined with Density Functional Theory (NEGF/DFT). The theoretical results
consider two different molecular orientations for each isomer. Whereas the orientation does not
affect the conductance of the trans isomer, it significantly modulates the conductance of the cis
isomer and the resulting conductance ON/OFF ratio of the molecular junction. This helps identifying
the molecular orientation at the origin of the observed current differences between the trans and cis
forms. The ON state is associated to the trans isomer irrespective of its orientation in the junction,
while the OFF state is identified as a cis isomer with its azobenzene moiety folded upward with
respect to the bithiophene core. The experimental and calculated ON/OFF conductance ratios have a
similar order of magnitude. This conductance ratio seems reasonable to make these Co-AzBT
molecular junctions a good test-bed to further explore the relationship between the spin-polarized
charge transport, the molecule conformation and the molecule-Co spinterface.
Keywords: molecular switch, electron transport, ferromagnetic electrode, self-assembly, spintronics,
NEGF/DFT.
�2
Page 4
INTRODUCTION
Molecular spintronics uses molecules and/or ensemble of molecules in devices in which the flow of
spin-polarized currents is used to manipulate and process information.1-4 Prototypical devices include
molecular magnetic tunnel junctions where a self-assembled monolayer (SAM) is used as a tunnel
layer between two ferromagnetic (FM) electrodes.5, 6 The study of photo-switchable molecular
junctions with FM electrodes is motivated by theoretical works7-9 predicting that the interplay
between molecular conformation and the injection of spin-polarized currents unlocks additional
degrees of freedom to control the spin-polarized transport in optically switchable molecular
spintronic devices. We have recently studied the optically induced conductance switching at the
nanoscale of diarylethene derivative SAMs on La0.7Sr0.3MnO3 electrodes,10 and observed a weak
conductance switching of the diarylethene molecular junctions (ON/OFF conductance ratios <8)
coupled to the optically induced resistive switching of the La0.7Sr0.3MnO3 substrate.
Here, we investigate the grafting of thiol-terminated molecular switches on Co surface under
controlled atmosphere and the electrical properties of the resulting Co-SAM/metal junctions (-
denotes here a covalent binding whereas / denotes a non-covalent binding) formed by conductive
atomic force microscopy (CAFM) under ultra-high vacuum (UHV), a mandatory condition to avoid
oxidation of the Co electrode. Indeed, despite previous realizations of molecular junctions employing
SAMs grafted on cobalt as bottom electrodes,11-17 electrical nanoscale characterization of SAMs on
cobalt remains limited, and photo-switchable molecular junctions on Co are not reported to date.
We study the properties of SAMs composed of a functional molecule, i.e. an azobenzene-
bithiophene molecular photoswitch (AzBT) (Scheme 1)18 possessing two photo-isomers, trans-AzBT
and cis-AzBT. The conversion from trans-AzBT to cis-AzBT takes place under UV irradiation
whereas the reverse reaction occurs under irradiation with blue light as usual for azobenzene
�3
Page 5
derivatives. Reversible isomerization was also demonstrated by plasmon excitation when AzBT
molecules are embedded in a network of gold nanoparticles.19 We have previously studied the
electronic properties of AzBT SAMs on gold by CAFM20 and in self-assembled networks of AzBT
functionalized gold nanoparticles.19, 21, 22 In all cases, the trans and cis isomers are respectively
associated with the OFF (less conducting) and ON (more conducting) states of the system. The
highest ON/OFF ratios (ratio of the currents in the cis and trans states at a given bias) are ~ 600 in
nanoparticles networks21 and ~103 with SAMs.20
We first characterize by various techniques (X-ray photoelectron spectroscopy, ellipsometry
and topographic AFM) the grafting of the thiol-terminated molecular switches on the surface of
cobalt under controlled atmosphere. We investigate the electrical properties (current vs. voltage, I-V,
curves) of the resulting Co-SAM/PtIr junctions formed by CAFM under ultra-high vacuum (UHV).
We found that, in all situations (pristine SAM, after UV and visible light irradiations), there is a
coexistence of the trans and cis azobenzene isomers in the SAMs. After UV light irradiation, the cis
isomer population increases while that of the trans isomer decreases. The respective populations are
reversed upon visible light irradiation. The conductance ratio between the two isomers is about 23 (at
0.5 V).
At the theoretical level, we perform a first-principle investigation of the electronic structure
and transport properties of molecular junctions including AzBT in several conformations
chemisorbed on the cobalt (111) surface. Since the work function modification upon SAM deposition
is related to the charge injection barriers from the metal to the active molecular layer,23-25 we start
our study by characterizing the change in the work function of the cobalt surface upon AzBT
deposition and then depict the energy level alignment at the Co-AzBT interface. In a next stage, we
simulate the transport properties of Co-AzBT junctions by means of the NEGF/DFT formalism.26
�4
Page 6
The transmission spectra at zero bias are calculated for the various molecular junctions and
rationalized by analyzing the nature of the transmitting molecular orbitals through the Local Device
Density of States (LDDoS) of the device. The I-V curves are obtained by using the Landauer-
Büttiker formalism within the coherent transport regime27 via integration of the transmission
spectrum in the bias window defined by the applied bias. The calculations identify the most probable
Co-AzBT interface conformations related to the experimentally observed changes in the molecular
junction conductance upon light irradiation. Based on our results, the high conductance ON state
corresponds to the trans conformation for the Co-AzBT interface irrespective of the orientation of
the azobenzene moiety with respect to the bithiophene unit (syn or anti), which gives almost the
same conductance. The lower conductance OFF state is associated with one cis conformation of the
Co-AzBT interface, with the azobenzene moiety folded upward. The lower conductance of the cis
conformation is attributed to an increase in both the hole injection barrier and tunneling barrier
length. We calculate ON/OFF (trans/cis) conductance ratios in good agreement with the
experimental data.
!
Scheme 1. Schematic representation of trans azobenzene-bithiophene (trans-AzBT, left) and cis
azobenzene-bithiophene (cis-AzBT, right) on Co electrodes connected by the CAFM tip (not to scale).
�5
Page 7
RESULTS
The synthesis of the AzBT molecules was reported elsewhere.18 The sample fabrication (SAMs on
evaporated Co substrates) was done in an inert atmosphere (glove box). The SAMs were formed
from a millimolar solution of AzBT in anhydrous toluene in the dark. For the transfer of the samples
between the globe box and various instruments (under inert atmosphere or UHV), we used a
hermetic container under an inert atmosphere (see Methods).
Structural characterization of the Co-AzBT SAMs
SAM morphology and thickness. The topographic images of the SAMs acquired simultaneously
with the CAFM images (see Methods) reveal a rather homogeneous surface free of pinholes with a
rms roughness of ~ 0.3 nm (Fig. 1a), and a spatially uniform distribution of low currents (~ few nA
at 0.5 V) (Fig. 1b). A first evaluation of the SAM thickness is obtained by the AFM nano-etching
technique (see Methods). The topographic images of the indented zone reveal the presence of a
square hole with a depth of ~1.1 nm (Figs. 1c and 1e, thickness averaged between the dashed lines
in Fig. 1c) where hot spots of current (up to 100 nA - 1 µA at 10 mV) are clearly observed (bright
spots in Fig. 1d) as well as a clear increase in the averaged current in the etched zone (Fig. 1f),
indicating the removal of the molecular layer. The thickness is confirmed by spectroscopic
ellipsometry (see Methods) on a freshly formed Co-AzBT SAM. We measure a thickness of 0.9 ± 0.2
nm, in good agreement with the value obtained by nano-etching. The thickness (ellipsometry)
increases to 1.5 ± 0.2 nm after trans-to-cis isomerization (UV light irradiation). Considering the
calculated (MOPAC, PM3 level)28 geometry optimized length of the molecules (3.1 nm for the trans
isomer, 2.6 nm for the cis isomer), the average tilt angle of the molecules in the SAM is estimated to
73 ± 4° with respect to the normal of the surface. We conclude that low-density SAMs (about 1.8
nm2/molecule, see theory section in the supporting information) are formed on the Co surface,
�6
Page 8
compared to SAMs of the same molecules on Au (SAM thickness 2.7 - 2.9 nm, with AzBT
molecules standing almost upright on the surface).20 It is likely that despite the strict protocol used
(see Methods), a weak oxidation of the Co substrate cannot be avoided (XPS section, vide infra),
preventing the formation of a densely packed SAM on the surface.
�7
Page 9
!
Figure 1. Topography (a) and log(current) (b) (1.0×1.0 µm2 area, F = 20 nN, V = +0.5 V) on Co-
AzBT. Scale bar length is 200 nm. Topography (c) and log(current) (d) (2.0×2.0 µm2, F = 20 nN, V =
+0.01 V). The squares indicate the location of a previous scan at F = 140 nN (nano-etching). Scale
bar length is 400 nm. (e, f) Respective mean sections of (c, d) calculated in the area delimited by
triangular cursors and dashed lines.
�8
Page 10
X-ray Photoemission Spectroscopy. The XPS spectrum of the freshly grafted SAM on cobalt surface
(Fig. 2a) shows the presence of: i) the different atoms of the AzBT molecule: carbon (C1s), nitrogen
(N1s), sulfur (S2p); ii) cobalt atoms from the substrate (Co2p); and iii) oxygen atoms (O1s). We also
record the Co2p, O1s and C1s spectra on an air-exposed cobalt substrate to compare the oxidation
state and the carbon contamination on the Co-AzBT sample (see the Supporting Information). The
Co2p spectra (Fig. 2b) shows the 2p3/2 and 2p1/2 peaks associated with metallic cobalt and cobalt
oxides. The first peak at 778.3 eV binding energy is in agreement with the work of Chuang et al.29
where this peak is observed at 778.2 eV and associated to the metallic Co2p3/2.30 The second peak at
780.5 eV corresponds to the binding energy of the cobalt oxides and hydroxides (CoO, Co3O4, Co2O3
CoOOH,...).29, 30 From the relative amplitudes of the peaks associated to cobalt oxides and
hydroxides with respect to the peaks associated to the metallic cobalt, the Co-AzBT sample shows a
larger amount of metallic cobalt than the air-exposed sample without SAM (Fig 2b). Albeit not
completely suppressing the Co oxidation, the glovebox environment, the protection by the SAM and
the hermetic homemade transport container clearly reduce the oxidation of the cobalt surface. The
shape of the Co2p spectra is similar to that of Co functionalized under glovebox by highly dense
SAMs of hexadecanethiol in the work of Hoertz et al..12 This presence of a residual oxidized Co in
the Co-AzBT is confirmed by the analysis of the O1s spectra (see the Supporting Information).
The C1s spectra (see the Supporting Information) can be deconvoluted in three peaks: the
main peak located at 285.1 eV is associated to the C-C and C-S bonds.31-33 The two other peaks with
binding energy 285.6 eV and 288.7 eV correspond to C-O and O-C=O bonds, respectively.34 These
last two peaks are not associated with the AzBT molecule but to a surface contamination. We
estimate a ratio of about 80/20 % for the carbon in AzBT and contaminants, respectively (see the
Supporting Information).
�9
Page 11
!
Figure 2. (a) XPS survey spectra of Co-AzBT (top) and of an air exposed cobalt substrate (bottom).
(b) High-resolution spectra of Co2p for Co-AzBT (top) and for air exposed cobalt substrate (bottom).
Dark grey peaks are associated to Co oxides and hydroxides and light grey peaks indicate the Co
metal. (c) High-resolution spectra of N1s for Co-AzBT. (d) High-resolution spectra of S2p for Co-
AzBT. Light grey peaks are associated to S bound to Co.
The peak position of the nitrogen N1s is observed at 400.2 eV (Fig. 2c), in good agreement
with previous results for AzBT grafted on gold substrate where the N1s peak was observed at 399.5
eV.20 This peak is clearly associated with the presence of the AzBT molecules.
The S2p spectrum (Fig. 2d) shows two doublets corresponding to the spin-orbit splitting of
the S2p level into a high intensity S2p3/2 peak at lower energy and a low intensity S2p1/2 peak at
higher energy. This doublet is separated by 1.2 eV with an intensity ratio of 2/1.18, 33, 35 Two doublets
�10
Page 12
are observed in the S2p spectrum of Co-AzBT SAM. The first doublet, with peaks located at 162.6
eV and 163.7 eV, corresponds to the S2p3/2 and S2p1/2 respectively. For the second doublet, the S2p3/2
peak is located at 165.4 eV and the S2p1/2 peak at 164.3 eV. By comparison, XPS of dodecanethiol
SAMs formed on cobalt substrate exhibits the S2p3/2 component associated to thiolate species
centered at 162.2 eV.14 This value is close to the S2p3/2 peak position of the first doublet at 162.6 eV
so that doublet may be assigned to the thiol chemisorbed on the cobalt surface (S2p3/2-Co). The
second doublet corresponds to other sulfur atoms in the body of the molecule (S2p3/2-C).
Electrical characterization by UHV Conductive-AFM.
The I-V curves are recorded with the UHV CAFM by forming a molecular junction with the PtIr tip
at a loading force of ~ 20 nN, resulting in a SAM deformation of ~0.2 nm, a contact area of ∼17
nm2 , i.e. ~10 molecules in the junction (see Methods and more details in the Supporting
Information). These curves are acquired just after the fabrication and transfer into the UHV CAFM.
Figure 3b shows the 2D I-V histogram of the pristine Co-AzBT/PtIr junctions constructed from 1200
I-V traces (400 I-V acquired on a 20 x 20 grid, pitch = 25 nm, repeated on 3 zones on the SAM, see
details in Methods and in the supporting information). These I-V curves have an almost symmetric
shape with respect to the bias polarity, as previously observed for the same molecules on Au
substrates.20 The 1D histograms taken at – 0.5 V and + 0.5 V (Figs. 3a and 3c) are decomposed into a
high and narrow peak at higher currents (HC peak) and a smaller and broader peak at lower currents
(LC peak), which are fitted by two log-normal distributions. Table 1 gives the fitted values of the
log-mean current (log-µ) and the log-standard deviation (log-σ). At both bias polarities, the LC and
HC peaks are centered in the nA range and separated by about a half-decade shift (HC peak: log-µ =
-8.38 (4.17 × 10-9 A), log-σ = 0.21 at -0.5 V, log-µ = -8.42 (3.80 × 10-9 A), log-σ = 0.19 at +0.5 V;
�11
Page 13
LC peak: log-µ = -8.84 (1.45 × 10-9 A), log-σ = 0.40 at -0.5 V, log-µ = -8.89 (1.29 × 10-9 A), log-σ =
0.39 at +0.5 V). From the areas of the two peaks, we deduce a relative contribution of 39% for the
HC peak and 61% for the LC one.
�12
Page 14
!
!
Figure 3. (a, c) Histograms and log-normal fits of log(current) at -0.5 V and +0.5 V. (b) 2D-
histogram of log(current) vs. voltage for Co-AzBT in the pristine state. (d-f) Same as (a-c) after 2h of
UV irradiation. Note that the horizontal red line between ca. -0.3 and 0.3V corresponds to the
preamplifier sensitivity limit of the current-voltage preamplifier (see supporting information) .(g-i)
Same as (a-c) after 2h of blue light irradiation. 1200 I-V traces are used in each histogram.
�13
Page 15
After 2 hours of UV irradiation (see Methods) to induce the trans-AzBT to cis-AzBT
isomerization, two peaks are still present in the statistical distribution (Fig. 3d-f). We observed two
main features; i) the global distribution is shifted to lower current values. At -0.5 and +0.5 V, the LC
peaks are now located in the 100 pA range while the HC peaks are slightly shifted and remain in the
nA range (HC peak: log-µ = -8.68 (2.09 × 10-9 A), log-σ = 0.16 at -0.5 V, log-µ = -8.76 (1.74 × 10-9
A), log-σ = 0.22 at +0.5 V, LC peak: log-µ = -9.65 (2.24 × 10-10 A), log-σ = 0.52 at -0.5 V, log-µ =
-9.79 (1.62 × 10-10 A), log-σ = 0.49 at +0.5 V). The centers of the HC and LC peaks are now
separated by about a decade; and ii) the asymmetry of the peak amplitudes is reversed. At both bias
polarities, the LC peak amplitude is higher than that of the HC peak, with a relative weight of 6-9%
for the HC peak and 91-94% for the LC peak (Table 1). Thus, after the UV irradiation, the Co-AzBT/
PtIr junctions have globally shifted to a lower conductance state. Irrespective to the switching
mechanism and the molecule conformation associated to the conductance states (see Discussion
section), we can quantify the overall conductance switching ratio of the Co-AzBT/PtIr junctions as
the ratio of the mean currents of the most intense, dominant, peaks in the pristine state (i.e., HC
peak) and after UV irradiation (i.e., LC peak). The highest ON/OFF conductance ratios are 19 and
23 at -0.5 and +0.5 V, respectively.
�14
Page 16
Table 1. Fitted parameters (log-normal) on the histograms of Fig. 3 for the LC (low conductance)
and HC (high conductance) peaks. Area under the peaks (a.u.) and relative weight of the peaks (%),
log-mean current (log-µ) and corresponding mean current (A),
and the log-standard deviation (log-σ).
This switching is reversible. After blue light irradiation (see Methods) to induce a cis-AzBT
to trans-AzBT isomerization, both the 2D I-V histograms and the 1D histograms at +/- 0.5 V (Fig.
3g-i) are almost similar as the pristine case. As in the pristine state, at both bias polarities the position
of the LC and HC peaks are in the nA range with a similar shift (about a half-decade) between them
(HC peak: log-µ = -8.54 (2.88 × 10-9 A), log-σ = 0.23 at -0.5 V, log-µ = -8.54 (2.88 × 10-9 A), log-σ =
0.19 at +0.5 V; LC peak: log-µ = -9.00 (10-9 A), log-σ = 0.45 at -0.5 V, log-µ = -9.05 (8.91 × 10-10 A),
log-σ = 0.50 at +0.5 V). The switch back is probably not complete under this light irradiation
condition since the HC/LC relative contributions are 24-31% for the HC peak and 69-76% for the
V = -0.5 V V = +0.5 VHC peak LC peak HC peak LC peak
PristineArea (relative contribution) 44.02 (39%) 69.77 (61%) 44.95 (39%) 68.90 (61%)log-µ (mean current (A)) -8.38 (4.17 × 10-9) -8.84 (1.45 × 10-9) -8.42 (3.80 × 10-9) -8.89 (1.29 × 10-9)log-σ 0.21 0.40 0.19 0.39
365 nm 2hArea (relative contribution) 6.56 (6%) 109.47 (94%) 10.45 (9%) 102.23 (91%)log-µ (mean current (A)) -8.68 (2.09 × 10-9) -9.65 (2.24 × 10-10) -8.76 (1.74 × 10-9) -9.79 (1.62 × 10-10)log-σ 0.16 0.52 0.22 0.49
470 nm 2hArea (relative contribution) 34.84 (31%) 76.95 (69%) 27.10 (24%) 85.65 (76%)log-µ (mean current (A)) -8.54 (2.88 × 10-9) -9.00 (10-9) -8.54 (2.88 × 10-9) -9.05 (8.91 × 10-10)
log-σ 0.23 0.45 0.19 0.50
�15
Page 17
LC peak, i.e. a HC peak amplitude slightly lower than for the pristine state (Table 1). Finally, given
the estimated number of molecules (~10, see above), we crudely infer a single molecule
conductance of 4.5x10-6G0/1.1x10-5G0 (min/max) for the HC peak and 4.1x10-7G0/3.7x10-6G0 (min/
max) for the LC peak (these values are calculated from the min and max of the mean values reported
in Table1, G0 is the quantum of conductance 7.75x10-5 S). We note that these estimates do not take
into account intermolecular interactions in the junctions which are known to significantly influence
the conductance of an ensemble of molecules in parallel.36, 37 We also note that the currents measured
for these Co-AzBT/PtIr junctions are lower than for a direct Co/PtIr contact (a factor of at least ~
103, see Fig. S1 in the supporting information), i.e. there is no short through the SAMs albeit the low
molecular packing in the monolayers.
Theoretical results.
We consider two Co-AzBT interfacial geometries for the trans and cis states of AzBT (Scheme 2).
For the trans state, we calculate the electronic and transport properties of two distinct conformations
where the azobenzene moiety is rotated by 180° (trans1 and trans2) with respect to the bithiophene
unit (syn- versus anti-orientation). For the cis state, we consider two extreme cases where the
azobenzene is folded upward (cis1) or downward (cis2). The molecules are tilted with respect to the
substrate to match the SAM thickness deduced experimentally for the trans and cis states,
corresponding to an area per molecule of 180 Å2 (see supporting information).
�16
Page 18
!
Scheme 2. Scheme of the four Co-AzBT conformations used for the calculations.
Co-AzBT self-assembled monolayers. Table 2 collects the work function shift computed for all
AzBT conformations together with the values of the molecular contribution ΔΦSAM and the bond
dipole ΔΦBD (see Methods and supporting information). The work function of the cobalt (111)
surface modified with the compounds in trans1, trans2, cis1 and cis2 conformations are 4.14 eV,
4.39 eV, 3.85 eV and 4.29 eV, respectively, which represents a decrease by -0.93 eV, -0.68 eV,
-1.22 eV and -0.78 eV with respect to the bare cobalt (111) surface (calculated to be 5.07 eV in
good agreement with the reported experimental value of 5.00 eV38). Thus, the cis1 (trans2) form
exhibits the higher (lower) work function shift. In order to elucidate the origin of this variation,
we compute the two components of the work function shift, namely the molecular contribution
(ΔΦSAM) and the bond dipole (ΔΦBD), as described in the methodology section and supporting
information. The results indicate that cis1 displays the higher molecular contribution of -0.74 eV
and a bond dipole of -0.48 eV. A similar value of the bond dipole (-0.47 eV) is obtained for the
trans2 conformation, which promotes, however, the smaller work function shift. The similar
amplitude of the bond dipole is explained by the fact that the isomerization of the azobenzene
part weakly perturbs the interfacial contact geometry; similar values of BD are also obtained for
the two other conformations. Thereby, the shift in the work function is dominated by the
�17
Page 19
difference in the dipole moment between the AzBT forms rather than by the electronic
reorganization upon formation of the Co-S bond. This is also evidenced by the magnitude of the
corresponding molecular dipole moment calculated according to the Helmholtz equation (Eq. S2
in the Supporting Information). In fact, when going from trans2 (the form with the smaller work
function shift) to cis1 (the form with the higher work function shift), the intrinsic dipole moment
µSAM evolves from 1.00 D to 3.52 D, which corresponds to a significant increase of 2.67 D,
whereas the dipole moment arising from the bond dipole, µBD, shows little variation (2.24 D and
2.28 D for trans2 and cis1, respectively).
Table 2. Calculated work function of Co-SAMs (Φ), work function shift (∆Φ) upon SAM
deposition, molecular contribution (∆ΦSAM), bond dipole (∆ΦBD), normal component of the
dipole moment (µSAM, µBD) obtained from the Helmholtz formula.
Besides the work function shift, the alignment of the frontier molecular orbitals with
respect to the energy of the Fermi level (EF) is another relevant interfacial parameter. Since the
HOMO and LUMO levels of the molecules are broadened and mixed with the continuum of
states of the cobalt surface, we compute (see Methods) their energy in the SAM by projecting the
density of states on the molecules, as displayed in figure 4. The HOMO and LUMO levels are
identified as the maximum of the broadened peaks. For all AzBT forms, the HOMO level is
closer to the Fermi level than the LUMO, implying that the HOMO level will provide the main
transport channel for charge flow through the molecular junctions (vide infra).
Φ (eV) ∆Φ (eV) ∆ΦSAM (eV) ∆ΦBD (eV) µSAM (D) µBD (D)
Co-trans1 4.14 -0.93 -0.36 -0.57 1.71 2.72
Co-trans2 4.39 -0.68 -0.21 -0.47 1.00 2.24
Co-cis1 3.85 -1.22 -0.74 -0.48 3.52 2.28
Co-cis2 4.29 -0.78 -0.18 -0.60 0.85 2.86
�18
Page 20
!
Figure 4. Projected Density of States (PDoS) on the AzBT molecules. The zero energy
is set to the Fermi energy.
Interestingly, trans1, trans2 and cis2 exhibit quite similar HOMO energies with respect to the
Fermi level for spin up (spin down) of -0.03 eV, -0.07 eV and -0.03 eV (-0.09 eV, -0.07 eV and
-0.09 eV) respectively, whereas cis1 displays a HOMO level at lower energy (-0.27 eV for both
spin up and spin down). Consistently, the increase in the energy barrier for hole injection (defined
here as EF-EHOMO) and the larger work function modification for cis1 will largely impact the
conductance of the corresponding molecular junction compared to the other configurations (vide
infra).
�19
Page 21
!
Figure 5. Optimized AzBT junctions. The calculated junction thickness is also marked. The cis1
form displays the larger thickness.
Zero-bias transmission. The zero-bias transmission spectra (calculated as the sum of spin-up and
spin-down transmission spectra) of the optimized Co-AzBT/Au junctions (Fig. 5) are calculated
and plotted in a linear scale in figure 6a (see Methods). Note that the choice of Au top electrodes
instead of a typical cobalt electrode is rationalized by two facts: (1) the measurements were
performed with a PtIr tip, which is not a ferromagnetic material; and (2) the work function
calculated for this top electrode is 4.81 eV, in good agreement with the experimental PtIr work
function of 4.86 eV.39 Many transmission peaks are located below the Fermi level within the
energy range from -1.50 to 0 eV while there is only one sharp transmission peak above the Fermi
level located at higher energy (about 1.68 eV). This reveals that occupied levels are responsible
for the significant transmission peaks involved as possible channels for charge transport through
the molecular junctions. For further explicit comparison, we plot in figure 6b the transmission
spectra of the Co-AzBT/Au junctions in a log scale within the energy range between -0.50 and
0.50 eV. Although cis1 exhibits a larger transmission band, the latter lies deeper with respect to
the Fermi level (-0.35 eV) whereas the two trans and the cis2 conformations display quite similar
transmission coefficients and peak energy positions.
�20
Page 22
!
Figure 6. (a) Linear and (b) log scale plot of the transmission spectra at zero bias (sum of spin-
up and spin-down transmission spectra) for all AzBT junctions. The dashed line refers to the
Fermi energy set to zero.
To reveal the origin of the peaks in the transmission spectra, the Local Device Density of States
(LDDoS), projected onto the real-space transport direction, is calculated at each transmission
peak energy (see Methods). As illustrated in figure 7, we clearly recognize the signature of the
HOMO shape characteristic of the isolated molecule for the dominant transmission peak of
trans1, trans2 and cis2 located at -0.11 eV, -0.15 eV, -0.13 eV, respectively. In contrast, the
LDDoS of the cis1 junction shows that the sharp peak at -0.35 eV matches the HOMO-1 shape of
the isolated molecule, rather than the HOMO level, which is located at a slightly higher energy
(-0.29 eV) with a lower transmission coefficient. This can be explained by the increase in the
junction thickness and hence the tunneling barrier length in cis1 that decouples the HOMO state
(a) (b)
�21
Page 23
from the top electrode (as indicated by a red arrow in Fig. 7) and thereby makes the HOMO-1
level, mainly localized on the right side of the molecule and filling most of the width of the
junction, as the source of the dominant transmission channel (Fig. 7). Overall, the lower
transmission coefficient at the Fermi energy of the cis1 conformation can thus be attributed to an
increase in both the injection barrier (EF-EHOMO) and the tunneling barrier length.
!
Figure 7. Shape of the transmitting occupied orbitals within the junction (LDDoS analysis) and
in gas phase (MO analysis). The LDDoS energies (at the peaks of T(E) in Fig. 6) are given with
respect to the Fermi level energy set to 0 eV.
Junctions under bias. The I-V curves of the Co-AzBT/Au junctions within the bias window
from 0 to 0.5 V were obtained from the Landauer-Büttiker formula27 by integration of the
transmission spectra at equilibrium within the bias window (defined by the magnitude of the
applied bias) by considering three extreme scenarios (see Methods and supporting information).
The first two scenarios consider that the voltage drop exclusively occurs at the weakest contact
�22
Page 24
(i.e., mechanical C-AFM tip contact) and that the Fermi level of gold top electrode is lowered by
-e⎪V⎪ for the positive voltage range and raised by e⎪V⎪ in the negative voltage range. In the third
scenario, the applied potential is symmetrically distributed, one Fermi level is raised by e⎪V⎪/2
and the other lowered by -e⎪V⎪/2. The results in Fig.S5 (supporting information) show that the
first and second scenario should give rise to a significant rectification of the I-V curve, which is
not seen experimentally. We thus conclude that the third scenario is more reasonable and is
therefore used to calculate the IV characteristics of the Co-AzBT/Au junctions. As shown in Fig.
8, the current for cis1 is distinctively lower than for trans and cis2 conformations. The I-V curves
of the trans isomers (trans1 and trans2) are quite similar, which indicates that the transport
properties of these two conformations are indiscernible.
!
Figure 8. Semilog scale plot of current (nA) versus bias (V) for the Co-AzBT/Au junctions.
Altogether, the simulation of Co-AzBT/Au junctions identifies two distinct states (trans and cis)
with different conductance. The ON-state is associated with the trans isomers while the OFF-
�23
Page 25
state is attributed to the signature of the cis1 isomer. To quantitatively evaluate the switching
effects between these two forms, we defined an ON/OFF ratio as I (trans) /I (cis1), where I
(trans) and I(cis1) represent the current through the trans1 or trans2 and cis1 forms, respectively.
We stress that the higher current slope that appears beyond 0.3V for cis1 (see Fig. 8) arises from
the integration of the equilibrium transmission spectrum, which is an approximation since any
shift of the electronic levels under bias is neglected. Since this feature is not seen in the
experimental curve, we speculate that the equilibrium transmission is not accurate beyond 0.3V
and hence that our theoretical ratios are more accurate below 0.3V. As displayed in Table 3, the
ON/OFF ratios in this low voltage are within the range of 25-44, which is in good agreement with
the experimentally measured ON/OFF conductance ratio of about 23 at 0.5 V (Fig. 3 and Table
1).
Table 3. The calculated ON/OFF ratio between trans and cis1 forms.
DISCUSSION
We clearly report the reversible, albeit partial, conductance switching of AzBT SAMs on Co. For the
pristine SAMs, we observe two peaks with a larger contribution of the HC peak. After UV irradiation
to induce the trans-to-cis isomerization, we observe a dominant contribution of the LC peak and a
strong decrease in the amplitude of the HC peak. Upon blue light irradiation, the initial situation (i.e.
the pristine case) is recovered. By comparison with the calculations, we conclude that the LC peak is
mainly due to AzBT in the cis1 conformation, which gives the lower transmission coefficient at EF
0.1V 0.2 V 0.3 V 0.4 V 0.5 V
Trans1/cis1 34.7 44.1 33.8 14.7 4.2
Trans2/cis1 26.6 24.9 15.9 6.3 1.7
�24
Page 26
(Fig. 6) and the lower conductance (Fig. 8). The HC peak is attributed to the AzBT in the trans
conformation, the two "model" cases (trans1 and trans2) showing similar properties, which are not
experimentally discernible according to theory (see Fig. 8) and given the broad current distribution
of the HC peak (Fig. 3). Moreover, it is likely that with the low surface coverage SAM, the
organizational disorder favors the coexistence of several conformations. A good agreement is
obtained between the experimental and calculated ON/OFF ratios.
This peak identification call for two remarks: i) For the Co-AzBT SAM, we have the
conductance Gcis < Gtrans. This feature is opposite to our previous observation for the same AzBT
molecules on gold (Gcis > Gtrans).20 This is explained by the difference in the SAM organization due
to lower density of the SAM on cobalt. On Au surfaces, the AzBT molecules in the trans isomer are
standing almost upright normal to the surface with the terminal CH3 acting as a tunnel barrier with
the CAFM tip, while for the cis isomer, the CAFM tip is directly contacting the N=N bond of the
azobenzene moiety, leading to a higher conductance Gcis.20 In addition, we note that upon trans-to-cis
isomerization, the AzBT SAM thickness decreases on Au, while it increases in the present case, thus
modulating the tunnel current accordingly. Several groups have reported Gcis > Gtrans in molecular
junctions including various azobenzene derivatives and electrodes (Au, graphene),40-46 though the
opposite situation, Gcis < Gtrans, was also reported.47-50 This discrepancy is not surprising since the
geometry of the junction at the atomic level is an important parameter. For example, the two cases
can be theoretically explained whether the molecule is chemisorbed at the two electrodes (Gcis <
Gtrans) or only at one electrode and physisorbed at the other one (Gcis > Gtrans).51 ii) The coexistence of
the trans and cis isomers in the pristine SAM is consistent with the composition in solution. From
UV-vis spectroscopy, we have previously observed two isosbestic points (at 297 nm and 380 nm)
related to the coexistence of the two isomers in equilibrium.18 Similarly, our previous study of AzBT
SAMs on Au electrodes showed a mixed population (about 55% cis and 45% trans) in the pristine
�25
Page 27
SAMs,20 consistent with the relative weights of the LC (cis isomer) and HC (trans isomer) peaks for
pristine SAMs on Co (Table 1).
After UV light irradiation, in addition to the change in their respective amplitudes, the two
LC and HC peaks are also globally shifted to lower currents (by a factor 3-8). This feature has to be
linked to the fact that the population of cis1 isomers increases in the SAMs. Thus, increasing the
population of the cis1 isomers in the SAM does increase the average hole barrier injection (EF-
EHOMO) at the Co-AzBT interface, as confirmed by the calculations. In addition, the increased
average SAM thickness globally reduces the current in the molecular junctions. This average
modification of the global electrostatic landscape and layer thickness induced by changes in the
population of the cis and trans isomers in the SAMs (Table 1) appears to be responsible for the
global shift of the measured current histograms (Fig. 3). Finally, the cis2 isomer has electronic
properties not discernible from those of the trans isomers. However, this conformation is less
probable because of the steric hindrance induced by the adjacent molecules that should prevent such
a downward folding of the azobenzene moiety. Moreover, the measurements clearly indicate a
significant modification of the conductance upon light irradiation (Fig. 3). Accordingly, cis2 has not
been considered in our analysis.
CONCLUSION
The functionalization of the cobalt surface with photochromic AzBT molecules is structurally
characterized and the measurements show the presence of a monolayer on the surface. The light-
induced isomerization of the molecules in the SAM induces clearly a reversible change in the
conductance histograms of the junctions with a ratio of about 23 between the currents at the
maximum of the distributions. By a careful comparison with first-principle calculations, we identify
�26
Page 28
the more conducting ON state to the trans isomer, while the cis isomer, upward folded, is responsible
for the less conducting OFF state. The calculated ON/OFF ratio is in good agreement with the
experiments. Albeit not gigantic, this ON/OFF ratio has a reasonable amplitude to make the present
Co-AzBT molecular junction a suitable test-bed to assess, at the nanoscale using an UHV CAFM
equipped with a magnetic field, the relationship between the spin-polarized electron transport, the
molecule conformation in the junction and the molecule-Co interface configurations, the molecule
orientations at a ferromagnetic electrode being an important factor of the spin-polarized electron
transport.52
METHODS
Sample fabrication.
General conditions of the process. To avoid the oxidation of the cobalt substrates, the fabrication of
the samples (i.e. cobalt deposition, preparation of the solutions, grafting of the SAMs and cleaning of
the samples) was carried out under nitrogen atmosphere inside a glovebox (MBRAUN) (H2O and O2
below 5 ppm). The glassware was oven dried at 120°C overnight before immediate insertion inside
the glovebox to remove residual adsorbed water on the surfaces. The different steps were quickly
sequenced to avoid downtime.
Cobalt substrate fabrication. We prepared cobalt substrates by evaporating about 40 nm of cobalt on
cleaved (12 ×10 mm) highly phosphorus-doped n-Si(100) substrates, resistivity < 5.10-3 Ω.cm (from
Siltronix), covered by native oxide, cleaned by 5 min sonication in acetone and isopropanol, then
rinsed with isopropanol and dried under N2 flow. The evaporation of 99.99% purity cobalt pellets
(Neyco) was realized by Joule effect in a vacuum evaporation system (Edwards Auto306) placed
inside the glovebox. The cobalt deposition was realized under a 10-6 mbar vacuum and at a low rate
deposition between 2 and 5 Å/s in order to minimize roughness.
�27
Page 29
Formation of SAMs on cobalt. The AzBT molecule was synthesized following a procedure
described elsewhere.18 AzBT SAMs were formed in anhydrous toluene (99.8% from Sigma Aldrich).
The solvent was stored over 4 Å molecular sieves (> 5 days), freshly activated (18 h at 250 °C) and
degassed with nitrogen (30 min). Freshly prepared cobalt substrates were immediately immersed in a
degassed millimolar solution of AzBT in anhydrous toluene for 22 h, in the dark, subsequently rinsed
with degassed anhydrous toluene and dried under N2 stream. Note that different samples, all prepared
following the same procedure, were used in every experiment carried out in this work, i.e. AFM
nano-etching, ellipsometry, XPS and UHV CAFM I-V spectroscopy.
Transfer under controlled atmosphere. To transfer the samples from the glovebox to the CAFM and
the X-ray Photoelectron Spectroscopy (XPS) instrument under UHV, the samples were attached on
holders for scanning probe microscopy (Scienta Omicron) by a metallic clamp to form an electrical
contact on the cobalt layer. The samples were transferred from the glovebox to the UHV instruments
in a homemade hermetic transport container under overpressure of N2 (atmosphere of the glovebox).
XPS analysis.
XPS experiments were performed to analyze the chemical composition of the SAMs and to check the
residual oxidation state of the cobalt surface. We used a Physical Electronics 5600 spectrometer
fitted in an UHV chamber with a residual pressure of 3x10-10 mbar. The measurements were done
using standard procedures (see details in the supporting information).
Spectroscopic ellipsometry.
The ellipsometry was used to determine the thickness of the Co-AzBT SAM. To avoid the oxidation
of the cobalt in contact with the atmosphere, the sample was placed in a sealed cell (HORIBA) filled
with the N2 atmosphere of the glovebox. This container was transferred to an UVISEL ellipsometer
(HORIBA) used to record spectroscopic ellipsometry data (see details the supporting information).
�28
Page 30
UHV CAFM measurements.
The CAFM experiments on SAMs under UHV (pressure 10-11 – 10-9 mbar) were performed at room
temperature using a VT-SPM microscope (Scienta Omicron). CAFM imaging and local current-
voltage (I-V) spectroscopy were carried out using Platinum-Iridium coated probes (SCM-PIC-V2
Bruker), the bias (V) was applied on the substrate. Typically, up to 1200 I-V traces were recorded at
a load force of ~ 20 nN to construct 2D I-V histograms (400 I-V recorded on a 20 x 20 grid, pitch =
25 nm, repeated on 3 zones on the SAM, see details in the supporting information). We also used
AFM nano-etching to estimate the SAM thickness. The SAM was intentionally indented applying a
high load force (140 nN) during a scan. Then, on an enlarged scan of the same zone, the topographic
images revealed the presence of a hole from which we estimated the SAM thickness.
Irradiation setup of the photochromic Co-AzBT SAMs.
We used a power LED (M365F1 from Thorlabs) for UV light irradiation. This LED has a wavelength
centered at 365 nm and a bandwidth of 7.5 nm. For the blue light irradiation, we used a power LED
(M470F1 from Thorlabs) which wavelength is centered at 470 nm with a bandwidth of 25 nm. In the
UHV CAFM, irradiation of the sample was performed in the entry lock of the instrument (P = 10-6
mbar N2). An optical fiber was brought close to the viewport. The measured power density at the
location of the sample in this configuration was 7.6 and 6.0 mW/cm² for the UV and blue light,
respectively.
Theoretical methodology.
To calculate the electronic structure and transport properties of Co-AzBT interfaces, different key
steps have been followed, as described below (more details in the supporting information).
Co-AzBT self-assembled monolayers. First, the geometries of the isolated AzBT molecules are
relaxed with forces below 0.01 eV/Å at the density functional theory (DFT) level using the Perdew-
�29
Page 31
Burke-Ernzerhof (PBE) functional within the generalized gradient approximation (GGA),53 as
implemented in the QuantumATK software (details in the supporting information).54, 55 The Co (111)
surface is modeled by a slab of five layers with 11 × 3 cobalt atoms (see figure S4 in the supporting
information). A vacuum region of approximately 30 Å is introduced above the surface and 10 Å
below it. The coordinates of the bottom three layers were kept fixed in the bulk lattice geometry
while the two top layers were relaxed until the final forces acting on the atoms are less than 0.02 eV/
Å. For this geometric relaxation, we use the Perdew-Burke-Ernzerhof (PBE) functional within the
spin generalized gradient approximation (SGGA),53, 56 a double-zeta plus polarization basis set, a
density mesh cutoff of 100 Ha and a (2×8×1) k-sampling. Once the geometry is relaxed, the
electronic structure and by extension the work function of the surface are computed using a 6×24×1
k-sampling, a 100 Ha mesh cutoff and a single-zeta plus polarization (SZP) basis set (see supporting
information). To build the Co-AzBT SAMs, the relaxed molecules are tilted to fit the measured SAM
thickness and then anchored on the cobalt surface through a sulfur atom. The SAM structures are
optimized by relaxing the molecules forming the SAMS and the top two metal layers until forces are
below 0.04 eV/ Å. The electronic structure and the SAM induced work function shift are then
computed following the same method as used for the calculations on the bare cobalt surface, with a
DZP basis set for the valence molecular orbitals (see details in the supporting information).
Co-AzBT molecular junctions. In a final stage, we add a second electrode on the top side of the
molecular layer to build a single heterogeneous molecular junction Co-AzBT/Au. This second
electrode is created by conserving the lattice parameters of the cobalt (111) surface and by
converting the cobalt atoms to gold atoms to avoid problems related to lattice incommensurability. A
van der Waals contact is assumed between the molecular layer and the top electrode, with an
interatomic distance determined as the sum of van der Waals radii of the hydrogen and gold atoms
(2.86 Å). The electronic transport calculations of the Co-AzBT/Au junctions were performed by the
�30
Page 32
combination of DFT to the Non-Equilibrium Green’s Function (NEGF) formalism,26 as implemented
in QuantumATK Q-2019.12-SP1 package (details are reported in the supporting information).54, 55
This formalism has been widely recognized as a robust tool to rationalize the experimental results
and predict new features and trends for charge transport in molecular junctions.57-59 Finally, the I-V
characteristics have been calculated on the basis of Landauer-Büttiker formalism,27 that links the
transmission spectrum to the current in a coherent transport regime (see the supporting information).
ASSOCIATED CONTENT
Supporting Information. This material is available free of charge via the Internet: Additional
characterization protocols (XPS, ellipsometry, CAFM) and theoretical methods, XPS spectra for
cobalt surface exposed to air and cobalt functionalized with AzBT SAM and tables with the fitting
parameters for XPS deconvolution.
AUTHOR CONTRIBUTIONS
L.T. and D.G. fabricated and characterized the SAMs. L.T. performed the UHV C-AFM
measurements. X.W. performed the XPS measurements. I.A. performed the theoretical calculations
under the supervision of C.v.D. and J.C. The experimental work done by L.T. is part of his PhD
thesis supervised by S.L. and T.M. The project was conceived by D.V. and managed by S.L. and D.V.
The manuscript was written by D.V. with the contributions of all the authors. All authors have given
approval of the final version of the manuscript.
# These authors (L.T. and I.A.) contributed equally to this work.
CONFLICT OF INTEREST
The authors declare no competing financial interest.
�31
Page 33
ACKNOWLEDGMENTS
This work has been financially supported by the French National Research Agency (ANR), project
SPINFUN ANR-17-CE24-0004 and partly by Renatech. We are grateful to the Philippe Blanchard’s
group at Moltech-Anjou (CNRS, U. Angers) for the AzBT synthesis. We acknowledge D. Deresmes
for his valuable help with the UHV CAFM instrument, J.L. Caudron, E. Galopin, L. Fugère for the
XPS measurements, Y. Deblock for ellipsometry, and J.-M. Mallet for the manufacturing of
mechanical parts of the transport container. The work of I.A. is supported by the Belgian National
Fund for Scientific Research (F.R.S.-FNRS) thanks to the project SPINFUN (Convention T.0054.20).
We also acknowledge the Consortium des Équipements de Calcul Intensif (CÉCI) funded by the
Belgian National Fund for Scientific Research (F.R.S.-FNRS) for providing the computational
resources. J.C. is an FNRS research director.
REFERENCES
1. Sanvito, S., Molecular spintronics. Chem. Soc. Rev. 2011, 40 (6), 3336.
2. Gu, H.; Zhang, X.; Wei, H.; Huang, Y.; Wei, S.; Guo, Z., An overview of the
magnetoresistance phenomenon in molecular systems. Chem. Soc. Rev. 2013, 42 (13), 5907.
3. Cinchetti, M.; Dediu, V. A.; Hueso, L. E., Activating the molecular spinterface. Nature
Materials 2017, 16 (5), 507-515.
4. Delprat, S.; Galbiati, M.; Tatay, S.; Quinard, B.; Barraud, C.; Petroff, F.; Seneor, P.;
Mattana, R., Molecular spintronics: the role of spin-dependent hybridization. J. Phys. D: Appl. Phys.
2018, 51 (47), 473001.
5. Petta, J. R.; Slater, S. K.; Ralph, D. C., Spin-dependent transport in molecular tunnel
junctions. Phys. Rev. Lett. 2004, 93 (13), 136601.
�32
Page 34
6. Wang, W.; Richter, C. A., Spin-polarized inelastic tunneling spectroscopy of a molecular
magnetic tunnel junction. Appl. Phys. Lett. 2006, 89, 153105.
7. Wang, Y.; Che, J. G.; Fry, J. N.; Cheng, H.-P., Reversible Spin Polarization at Hybrid
Organic–Ferromagnetic Interfaces. J. Phys. Chem. Lett. 2013, 4 (20), 3508-3512.
8. Ulman, K.; Narasimhan, S.; Delin, A., Tuning spin transport properties and molecular
magnetoresistance through contact geometry. The Journal of Chemical Physics 2014, 140 (4),
044716.
9. Zeng, J.; Chen, K.-Q.; Deng, X.; Long, M., Light-driven strong spin valve effects in an
azobenzene-based spin optoelectronic device. J. Phys. D: Appl. Phys. 2016, 49 (41), 415104.
10. Thomas, L.; Guerin, D.; Quinard, B.; Jacquet, E.; Mattana, R.; Seneor, P.; Vuillaume, D.;
Mélin, T.; Lenfant, S., Conductance switching at the nanoscale of diarylethene derivative self-
assembled monolayers on La 0.7Sr 0.3MnO 3. Nanoscale 2020, 12, 8268-8276.
11. Fan, X.; Rogow, D. L.; Swanson, C. H.; Tripathi, A.; Oliver, S. R. J., Contact printed Co/
insulator/Co molecular junctions. Appl. Phys. Lett. 2007, 90 (16), 163114.
12. Hoertz, P. G.; Niskala, J. R.; Dai, P.; Black, H. T.; You, W., Comprehensive investigation of
self-assembled monolayer formation on ferromagnetic thin film surfaces. J. Am. Chem. Soc. 2008,
130 (30), 9763-9772.
13. Niskala, J. R.; You, W., Metal−Molecule−Metal Junctions via PFPE Assisted Nanotransfer
Printing (nTP) onto Self-Assembled Monolayers. J Am Chem Soc 2009, 131 (37), 13202-13203.
14. Devillers, S.; Hennart, A.; Delhalle, J.; Mekhalif, Z., 1-Dodecanethiol Self-Assembled
Monolayers on Cobalt. Langmuir 2011, 27, 14849-14860.
15. Catarelli, S. R.; Higgins, S. J.; Schwarzacher, W.; Mao, B.-W.; Yan, J.-W.; Nichols, R. J.,
Ionic liquid based approach for single-molecule electronics with cobalt contacts. Langmuir 2014, 30
(47), 14329-14336.
�33
Page 35
16. Galbiati, M.; Delprat, S.; Mattera, M.; Mañas-Valero, S.; Forment-Aliaga, A.; Tatay, S.;
Deranlot, C.; Seneor, P.; Mattana, R.; Petroff, F., Recovering ferromagnetic metal surfaces to fully
exploit chemistry in molecular spintronics. AIP Advances 2015, 5 (5), 057131.
17. Pookpanratana, S.; Lydecker, L. K.; Richter, C. A.; Hacker, C. A., Self-Assembled
Monolayers Impact Cobalt Interfacial Structure in Nanoelectronic Junctions. J. Phys. Chem. C 2015,
119, 6687-6695.
18. Karpe, S.; Ocafrain, M.; Smaali, K.; Lenfant, S.; Vuillaume, D.; Blanchard, P.; Roncali, J.,
Oligothiophene-derivatized azobenzene as immobilized photoswitchable conjugated systems. Chem
Commun 2010, 46 (21), 3657-3659.
19. Stievenard, D.; Guerin, D.; Lenfant, S.; Leveque, G.; Nijhuis, C. A.; Vuillaume, D.,
Electrical detection of plasmon-induced isomerization in molecule-nanoparticle network devices.
Nanoscale 2018, 10, 23122–23130.
20. Smaali, K.; Lenfant, S.; Karpe, S.; Oçafrain, M.; Blanchard, P.; Deresmes, D.; Godey, S.;
Rochefort, A.; Roncali, J.; Vuillaume, D., High On−Off Conductance Switching Ratio in Optically-
Driven Self-Assembled Conjugated Molecular Systems. ACS Nano 2010, 4 (4), 2411-2421.
21. Viero, Y.; Copie, G.; Guerin, D.; Krzeminski, C.; Vuillaume, D.; Lenfant, S.; Cleri, F.,
High Conductance Ratio in Molecular Optical Switching of Functionalized Nanoparticle Self-
Assembled Nanodevices. J. Phys. Chem. C 2015, 119, 21173-21183.
22. Viero, Y.; Guerin, D.; Alibart, F.; Lenfant, S.; Vuillaume, D., Light-stimulable molecules/
nanoparticles multi-terminal networks for switchable logical function and reservoir computing. Adv.
Func. Mater. 2018, 28, 1801506.
23. Heimel, G.; Romaner, L.; Brédas, J.-L.; Zojer, E., Organic/metal interfaces in self-
assembled monolayers of conjugated thiols: A first-principles benchmark study. Surface Science
2006, 600 (19), 4548-4562.
�34
Page 36
24. Heimel, G.; Rissner, F.; Zojer, E., Modeling the Electronic Properties of π-Conjugated Self-
Assembled Monolayers. Adv. Mater. 2010, 22 (23), 2494-2513.
25. Osella, S.; Cornil, D.; Cornil, J., Work function modification of the (111) gold surface
covered by long alkanethiol-based self-assembled monolayers. Physical Chemistry Chemical Physics
2014, 16 (7), 2866-2873.
26. Brandbyge, M.; Mozos, J.-L.; Ordejón, P.; Taylor, J.; Stokbro, K., Density-functional
method for nonequilibrium electron transport. Phys. Rev. B 2002, 65 (16), 165401.
27. Büttiker, M.; Imry, Y.; Landauer, R.; Pinhas, S., Generalized many-channel conductance
formula with application to small rings. Physical Review B 1985, 31 (10), 6207-6215.
28. MOPAC CS MOPAC, CambridgeSoft Corporation: Cambridge, 1996.
29. Chuang, T. J.; Brundle, C. R.; Rice, D. W., Interpretation of the x-ray photoemission spectra
of cobalt oxides and cobalt oxide surfaces. Surface Science 1976, 59 (2), 413-429.
30. Moulder, J. F.; Chastain, J., Handbook of X-ray photoelectron spectroscopy: A reference book
of standard spectra for identification and interpretation of XPS data. Perkin-Elmer Corporation:
1992.
31. Beulen, M. W. J.; Kastenberg, M. I.; van Veggel, F. C. J. M.; Reinhoudt, D. N.,
Electrochemical stability of self-assembled monolayers on gold. Langmuir 1998, 14 (26),
7463-7467.
32. Bourg, M. C.; Badia, A.; Lennox, R. B., J. Phys. Chem. B 2000, 104, 6562.
33. Vance, L.; Willey, T. M.; Nelson, A. J.; van Buuren, T.; Bostedt, C.; Terminello, L. J.; Fox,
J. A., Langmuir 2002, 18, 8123.
34. Stipp, S. L.; Hochella, M. F., Structure and bonding environments at the calcite surface as
observed with X-ray photoelectron spectroscopy (XPS) and low energy electron diffraction (LEED).
Geochimica et Cosmochimica Acta 1991, 55 (6), 1723-1736.
�35
Page 37
35. Lenfant, S.; Viero, Y.; Krzeminski, C.; Vuillaume, D.; Demeter, D.; Dobra, I.; Ocafrain,
M.; Blanchard, P.; Roncali, J.; Van Dyck, C.; Cornil, J., New Photomechanical Molecular Switch
Based on a Linear pi-Conjugated System. Journal of Physical Chemistry C 2017, 121 (22),
12416-12425.
36. Reuter, M. G.; Hersam, M. C.; Seideman, T.; Ratner, M. A., Signatures of Cooperative
Effects and Transport Mechanisms in Conductance Histograms. Nano Lett 2012, 12 (5), 2243-2248.
37. Trasobares, J.; Rech, J.; Jonckheere, T.; Martin, T.; Aleveque, O.; Levillain, E.; Diez-
Cabanes, V.; Olivier, Y.; Cornil, J.; Nys, J. P.; Sivakumarasamy, R.; Smaali, K.; Leclere, P.;
Fujiwara, A.; Theron, D.; Vuillaume, D.; Clement, N., Estimation of π-π Electronic Couplings from
Current Measurements. Nano Lett 2017, 17 (5), 3215–3224.
38. Michaelson, H. B., The work function of the elements and its periodicity. J. Appl. Phys. 1977,
48 (11), 4729-4733.
39. Lee, N. J.; Yoo, J. W.; Choi, Y. J.; Kang, C. J.; Jeon, D. Y.; Kim, D. C.; Seo, S.; Chung, H.
J., The interlayer screening effect of graphene sheets investigated by Kelvin probe force microscopy.
Appl. Phys. Lett. 2009, 95 (22), 222107.
40. Ferri, V.; Elbing, M.; Pace, G.; Dickey, M. D.; Zharnikov, M.; Samorì, P.; Mayor, M.;
Rampi, M. A., Light-Powered Electrical Switch Based on Cargo-Lifting Azobenzene Monolayers.
Angew. Chem. Int. Ed. 2008, 47 (18), 3407-3409.
41. Mativetsky, J. M.; Pace, G.; Elbing, M.; Rampi, M. A.; Mayor, M.; Samori, P.,
Azobenzenes as Light-Controlled Molecular Electronic Switches in Nanoscale Metal-Molecule-
Metal Junctions. J. Am. Chem. Soc. 2008, 130 (29), 9192-9193.
42. Faramarzi, V.; Raimondo, C.; Reinders, F.; Mayor, M.; Samorì, P.; Doudin, B., Optically
switchable molecular device using microsphere based junctions. Appl. Phys. Lett. 2011, 99 (23),
233104.
�36
Page 38
43. Kim, Y.; Wang, G.; Choe, M.; Kim, J.; Lee, S.; Park, S.; Kim, D.-Y.; Lee, B.; Lee, T.,
Electronic properties associated with conformational changes in azobenzene-derivative molecular
junctions. Organic Electronics 2011, 12 (12), 2144-2150.
44. Seo, S.; Min, M.; Lee, S. M.; Lee, H., Photo-switchable molecular monolayer anchored
between highly transparent and flexible graphene electrodes. Nature Communications 2013, 4 (1),
1920.
45. Margapoti, E.; Strobel, P.; Asmar, M. M.; Seifert, M.; Li, J.; Sachsenhauser, M.; Ceylan,
Ö.; Palma, C.-A.; Barth, J. V.; Garrido, J. A.; Cattani-Scholz, A.; Ulloa, S. E.; Finley, J. J.,
Emergence of Photoswitchable States in a Graphene–Azobenzene–Au Platform. Nano Lett 2014, 14
(12), 6823-6827.
46. Margapoti, E.; Li, J.; Ceylan, Ö.; Seifert, M.; Nisic, F.; Anh, T. L.; Meggendorfer, F.;
Dragonetti, C.; Palma, C.-A.; Barth, J. V.; Finley, J. J., A 2D Semiconductor–Self-Assembled
Monolayer Photoswitchable Diode. Adv. Mater. 2015, 27 (8), 1426-1431.
47. Wen, Y.; Yi, W.; Meng, L.; Feng, M.; Jiang, G.; Yuan, W.; Zhang, Y.; Gao, H.; Jiang, L.;
Song, Y., Photochemical-Controlled Switching Based on Azobenzene Monolayer Modified Silicon
(111) Surface. The Journal of Physical Chemistry B 2005, 109 (30), 14465-14468.
48. Kumar, A. S.; Ye, T.; Takami, T.; Yu, B.-C.; Flatt, A. K.; Tour, J. M.; Weiss, P. S.,
Reversible Photo-Switching of Single Azobenzene Molecules in Controlled Nanoscale
Environments. Nano Lett. 2008, 8 (6), 1644-1648.
49. Ely, T.; Das, S.; Li, W.; Kundu, P. K.; Tirosh, E.; Cahen, D.; Vilan, A.; Klajn, R.,
Photocontrol of Electrical Conductance with a Nonsymmetrical Azobenzene Dithiol. Synlett 2013,
24 (18), 2370-2374.
50. Döbbelin, M.; Ciesielski, A.; Haar, S.; Osella, S.; Bruna, M.; Minoia, A.; Grisanti, L.;
Mosciatti, T.; Richard, F.; Prasetyanto, E. A.; De Cola, L.; Palermo, V.; Mazzaro, R.; Morandi,
V.; Lazzaroni, R.; Ferrari, A. C.; Beljonne, D.; Samorì, P., Light-enhanced liquid-phase exfoliation
�37
Page 39
and current photoswitching in graphene–azobenzene composites. Nature Communications 2016, 7
(1), 11090.
51. Osella, S.; Samorì, P.; Cornil, J., Photoswitching Azobenzene Derivatives in Single Molecule
Junctions: A Theoretical Insight into the I/ VCharacteristics. J. Phys. Chem. C 2014, 118,
18721-18729.
52. Brooke, R. J.; Jin, C.; Szumski, D. S.; Nichols, R. J.; Mao, B.-W.; Thygesen, K. S.;
Schwarzacher, W., Single-Molecule Electrochemical Transistor Utilizing a Nickel-Pyridyl
Spinterface. Nano Lett 2014, 15 (1), 275-280.
53. Perdew, J. P.; Burke, K.; Ernzerhof, M., Generalized Gradient Approximation Made Simple.
Phys. Rev. Lett. 1996, 77 (18), 3865-3868.
54. QuantumATK Q-2019.12-SP1, Synopsis QuantumATK, www.synopsys.com/silicon/
quantumatk.html.
55. Smidstrup, S.; Stradi, D.; Wellendorff, J.; Khomyakov, P. A.; Vej-Hansen, U. G.; Lee, M.-
E.; Ghosh, T.; Jónsson, E.; Jónsson, H.; Stokbro, K., First-principles Green's-function method for
surface calculations: A pseudopotential localized basis set approach. Physical Review B 2017, 96
(19), 195309.
56. Smidstrup, S.; Markussen, T.; Vancraeyveld, P.; Wellendorff, J.; Schneider, J.; Gunst, T.;
Verstichel, B.; Stradi, D.; Khomyakov, P. A.; Vej-Hansen, U. G.; Lee, M.-E.; Chill, S. T.;
Rasmussen, F.; Penazzi, G.; Corsetti, F.; Ojanperä, A.; Jensen, K.; Palsgaard, M. L. N.; Martinez,
U.; Blom, A.; Brandbyge, M.; Stokbro, K., QuantumATK: an integrated platform of electronic and
atomic-scale modelling tools. Journal of Physics: Condensed Matter 2019, 32 (1), 015901.
57. Schmaus, S.; Bagrets, A.; Nahas, Y.; Yamada, T. K.; Bork, A.; Bowen, M.; Beaurepaire,
E.; Evers, F.; Wulfhekel, W., Giant magnetoresistance through a single molecule. Nature Nanotech.
2011, 6, 185-189.
�38
Page 40
58. Van Dyck, C.; Geskin, V.; Kronemeijer, A. J.; De Leeuw, D. M.; Cornil, J., Impact of
derivatization on electron transmission through dithienylethene-based photoswitches in molecular
junctions. Phys. Chem. Chem. Phys. 2013, 15 (12), 4392-4404.
59. Rodriguez-Gonzalez, S.; Xie, Z.; Galangau, O.; Selvanathan, P.; Norel, L.; Van Dyck, C.;
Costuas, K.; Frisbie, C. D.; Rigaut, S.; Cornil, J., HOMO Level Pinning in Molecular Junctions:
Joint Theoretical and Experimental Evidence. J. Phys. Chem. Lett. 2018, 9 (9), 2394-2403.
�39
Page 41
Conductance Switching of Azobenzene-Based Self-Assembled
Monolayers on Cobalt Probed by UHV Conductive-AFM.
Louis Thomas,1,# Imane Arbouch,2,# David Guérin,1 Xavier Wallart,1 Colin van Dyck,2 Thierry Mélin,1
Jérôme Cornil,2,* Dominique Vuillaume1,* and Stéphane Lenfant1,*
1) Institute of Electronics Microelectronics and Nanotechnology (IEMN), CNRS, University of Lille,
Avenue Poincaré, Villeneuve d'Ascq, France.
2) Laboratory for Chemistry of Novel Materials, University of Mons, Place du Parc 20, Mons,
Belgium.
Supporting Information
1. XPS characterization.
High resolution XPS spectra were recorded with a monochromatic AlKα X-ray source (hυ = 1486.6
eV), a detection angle of 45° as referenced to the sample surface, an analyzer entrance slit width of 400
µm and with an analyzer pass energy of 12 eV. In these conditions, the overall resolution as measured
from the full-width half-maximum (FWHM) of the Ag 3d5/2 line is 0.55 eV. Background was subtracted
by the Shirley method.1 The peaks were decomposed using Voigt functions and a least squares
minimization procedure.
2. Ellipsometry.
The system acquired a spectrum ranging from 2 to 4.5 eV (corresponding to 300 to 750 nm) with
intervals of 0.1 eV (or 15 nm). To use the cell filled with N2, data were taken at an angle of incidence of
�1
Page 42
60 ± 1° while the compensator was set at 45°. We fitted the data with DeltaPsi 2 data analysis software
by a regression analysis to a film on a substrate model as described by their thickness and their
complex refractive indexes. First, we recorded a background before monolayer deposition for a cobalt
substrate freshly evaporated on the silicon substrate. We acquired three reference spectra at three
different places of the surface spaced of few mm. Secondly, after the monolayer deposition, we
acquired once again three spectra at three different places of the surface. We used a 2-layer model
(substrate/SAM) to fit the measured data and to determine the SAM thickness. We used the previously
measured optical properties of the cobalt substrate (background) and we fixed the refractive index of
the organic monolayer at 1.50. The usual values in the literature for the refractive index of organic
monolayers are in the range 1.45-1.50.2, 3 The three spectra measured on the sample were fitted
separately using each of the three reference spectra, giving nine values for the SAM thickness. We
calculated the mean value from this nine thickness values and the thickness incertitude corresponding
to the standard deviation.
3. CAFM measurements.
CAFM imaging and local current-voltage (I-V) spectroscopy were carried out using Platinum-Iridium
coated probes SCM-PIC-V2 (Bruker), tip radius R = 25 nm, spring constant k = 0.1 N/m. In all our
measurements, bias (V) was applied on the substrate and the current was recorded with an external
DLPCA-200 amplifier (FEMTO) at a gain of 1µA/V, which set the current sensitivity limit at ca. 10-10
A. The force applied on the sample by the tip was calculated from approach-retract curves prior to each
experiment to ensure the correct vertical deflection set point with respect to the fixed force set point
(F). The load force was set at ~ 20 nN for all the I-V measurements, a lower value leading to too many
contact instabilities during the I-V measurements. Albeit larger than the usual load force (2-5 nN) used
�2
Page 43
for CAFM on SAMs, this value is below the limit of about 60-70 nN at which the SAMs start to suffer
from severe degradations. For example, a detailed study (Ref. 4) showed a limited strain-induced
deformation of the monolayer (≲ 0.3 nm) at this used load force. The same conclusion was confirmed
by our own study comparing mechanical and electrical properties of alkylthiol SAMs on flat Au
surfaces and tiny Au nanodots.5 Moreover, we checked by topographic AFM that a load force of 20 nN
is not indenting the SAM.6 Current-voltage spectra (I-V) were acquired on several grids (20 x 20 points
spaced by 25 nm from each other), each grid spaced of a few mm. Data were processed using
Gwyddion7 and WSxM8 software. We typically recorded 1200 I-V traces, used without data selection to
construct 2D histogram of the decimal logarithm of the current versus voltage. For a given bias, we
fitted the current histogram with log-normal distributions and extracted the relevant statistical
quantities: the log-mean current (log-µ) and the log-standard deviation (log-σ). Topographic and
current images (at a given bias) were also recorded simultaneously. As usually reported in literature4,
9-11 the contact radius (a) between the C-AFM tip and the SAM surface, and the SAM elastic
deformation (δ) are estimated from a Hertzian model:12
% (S1)
% (S2)
with F the tip load force (20 nN), R the tip radius (25 nm) and E* the reduced effective Young modulus
defined as:
% (S3)
a2 = 3RF4E *
⎛⎝⎜
⎞⎠⎟
2/3
δ = 916R
⎛⎝⎜
⎞⎠⎟
1/3FE *
⎛⎝⎜
⎞⎠⎟
2/3
E * = 1ESAM* + 1
Etip*
⎛
⎝⎜⎜
⎞
⎠⎟⎟
−1
=1−ν SAM
2
ESAM+1−νtip2Etip
⎛
⎝⎜⎜
⎞
⎠⎟⎟
−1
�3
Page 44
In this equation, ESAM/tip and νSAM/tip are the Young modulus and the Poisson ratio of the SAM and C-
AFM tip, respectively. For the Pt/Ir (90%/10%) tip, we have Etip = 204 GPa and νSAM/tip = 0.37 using a
rule of mixture with the known material data.13 These parameters for the AzBT SAM are not known
and, in general, they are not easily determined in such a monolayer material. Thus, we consider the
value of an effective Young modulus of the SAM E*SAM = 38 GPa as determined for the "model
system" alkylthiol SAMs from a combined mechanics and electron transport study.4 With these
parameters, we estimate a = 2.3 nm (contact area = 16.6 nm2) and δ = 0.2 nm. With a molecular
packing density of 1.8 nm2/molecule (see theory, below §6), we infer that about 10 molecules are
measured in the Co-AzBT/PtIr junction.
We also checked that there is no short through the SAMs by comparing the I-Vs for the Co-
AzBT/PtIr tip (Fig. 3, main text) with the typical I-Vs for a direct contact of the C-AFM tip on Co
surface (Fig. S1), which show a current reaching the compliance limit of the preamplifier (here 10 µA)
at V < 50 mV.
%
Figure S1. Typical I-Vs (20 traces) for the direct contact of the PrIr C-AFM tip on a Co surface.
�4
Page 45
4. XPS analysis of air-exposed cobalt
%
Figure S2. XPS spectra of air-exposed cobalt: (a) survey, (b) C1s, (c) O1s, (d) Co2p.
Table S1. Identification of the peaks in the XPS analysis of air-exposed cobalt.
Peak assignation from Refs. 14-18
Spectra Identification Energy (eV)
C1s C-C 285.2
C1s C-O-C 286.3
C1s C-C=O 289.3
O1s CoO, Co3O4 529.8
O1s C-O, Co(OH)2 531.8
Co2p Co metal 2p3/2 778.2
Co2p CoO, Co3O4, Co(OH)2 2p3/2 780.3
Co2p satellite 785.8
Co2p Co metal 2p1/2 793.4
Co2p CoO, Co3O4, Co(OH)2 2p1/2 796.4
Co2p satellite 802.6
�5
Page 46
5. XPS analysis of Co-AzBT SAM
The O1s spectra is composed of two peaks at binding energies of 529.7 eV and 531.6 eV. These values
are close to those observed by Chuang et al.19 on cobalt surface exposed to air (529.5 eV and 530.8
eV). For these authors these peaks are associated to the oxidized cobalt. From the survey spectra (Fig.
S3a), a smaller amount of oxygen is detected on Co-AzBT than on the air-exposed sample (Fig. S2a).
For the Co-AzBT XPS analysis (Table S2), C1s peaks were observed approximately at the same
�6
�
Figure S3. XPS spectra of Co-AzBT: (a) survey, (b) C1s, (c) O1s, (d) Co2p, (e) S2p, (f) N1s.
Page 47
binding energy as on a cobalt surface exposed to the air without SAM (see Table S1). The first peak
associated to C-C and C-S bonds and located at 285.1 eV was also observed on the sample of air-
oxidized cobalt without SAM at 285.2 eV. However, the peak area is lower for the air-oxidized cobalt
(divided by 4.8) compared to the sample with the SAM (Co-AzBT), with a peak area of 5192 (a.u.) and
1087 (a.u.) for Co-AzBT and air-oxidized cobalt respectively. So, the peak at 285.1 eV observed on
Co-AzBT is a contribution of the carbon atoms of the AzBT molecule but also of the surface
contamination. If we suppose that the contamination is the same for both the samples, we estimate that
the peak observed at 285.1 eV on Co-AzBT is due to 21 % to the contamination and 79 % to the AzBT
molecule grafted on the cobalt surface.
�7
Page 48
Table S2. Identification of the peaks in the XPS analysis of Co-AzBT.
Peak assignation from Refs. 14-18
6. Full theoretical methodology.
Co-AzBT self-assembled monolayers. First, the geometries of the isolated AzBT molecules are relaxed
with forces below 0.01 eV/Å at the density functional theory (DFT) level using the Perdew-Burke-
Ernzerhof (PBE) functional within the generalized gradient approximation (GGA),20 as implemented in
the QuantumATK software.21, 22 The valence electrons are described within the LCAO approximation,
with a double-zeta plus polarization basis set (DZP) whilst the core electrons are described by the
norm-conserving Troullier-Martins pseudopotentials.23 We use a density mesh cutoff of 100 Ha and
Spectra Peak Identification Energy (eV)
C1s C-C, C-S 285.1
C1s C-O-C 285.6
C1s C-C=O 288.7
O1s CoO, Co3O4 529.7
O1s C-O, Co(OH)2 531.6
Co2p Co metal 2p3/2 778.3
Co2p CoO, Co3O4 2p3/2 780.5
Co2p satellite 785.1
Co2p Co metal 2p1/2 793.4
Co2p CoO, Co3O4 2p1/2 796.4
Co2p satellite 802.6
S2p e1' CoS S2p3/2 162.6
S2p e2' CoS S2p1/2 163.7
S2p e3' CS S2p3/2 164.3
S2p e4' CS S2p1/2 165.4
N1s N in azobenzene 400.2
�8
Page 49
multipole boundary conditions to correct for the interaction between the single molecules introduced in
a large unit cell and their images generated by the Periodic Boundary Conditions (PBC). The Co(111)
surface is modeled by a slab of five layers with 11 × 3 cobalt atoms in each layer and lattice parameters
a= 27.645 Å, b= 7.539 Å and α= 120°. This corresponds to a theoretical area per molecule of 180 Å2
(see Fig. S4). A vacuum region of approximately 30 Å was introduced above the surface and 10 Å
below it. The coordinates of the bottom three layers were kept fixed in the bulk lattice geometry while
the two top layers were relaxed until the final forces acting on the atoms are less than 0.02 eV/Å. For
this relaxation, we use the Perdew-Burke-Ernzerhof (PBE) functional with the spin generalized
gradient approximation (SGGA),20, 24 a double-zeta plus polarization basis set, a density mesh cutoff of
100 Ha and a (2×8×1) k-sampling.
%
Figure S4. (a) side and (b) top view of the Co (111) slab.
Once the geometry is relaxed, the electronic structure and by extension the work function of the
surface are computed using a 6×24×1 k-sampling, a 100 Ha mesh cutoff and a single-zeta plus
polarization (SZP) basis set. We reduce the basis set here to be consistent with electronic structure
computations achieved on larger systems such as SAMs and molecular junctions. For the resolution of
Poisson’s equation, Neumann (fixed potential gradient) and Dirichlet (fixed potential) were imposed on
the left- and right-hand side of the slab, respectively, combined with periodic boundary conditions
(PBC) in the in-plane directions. These mixed PBCs provide an alternative to the widely used slab
�9
Page 50
dipole correction.25, 26 The Fermi energy is calculated initially for cobalt so that its energetic position
ensures that the occupation of the density of states of the system equals the total number of electrons.
Within this framework, we compute a work function of 3.78 eV for the Co(111) surface, a value too
low compared to the experimental value of 5.00 eV.27 This result is fully expected due to the use of
short range localized atomic orbitals that induces an artificial push-back effect at the interface and thus
reduces the metal work function. To solve this numerical problem, we added a layer of ghost atoms in
the vacuum next to the surface to better describe the tail of the electron density, an approach already
implemented in QuantumATK Q-2019.12-SP1.21 Several tests have been performed in order to identify
the nature of the ghost atoms, the converged distance from the top layer, the basis set and the
pseudopotential that provide good accuracy at reasonable computational times. The obtained results
indicate that the use of a layer of platinum ghost atoms described by medium basis sets and
PseudoDojo pseudopotentials28 at a distance of 1.6 Å from the top cobalt layer provides a work
function of 5.07 eV, which is in very good agreement with the reported experimental value of 5.00 eV.27
To build the Co-AzBT SAMs, the molecules are tilted to fit the measured SAM thickness and
then anchored on the cobalt surface through a sulfur atom. The SAM structures are optimized by
relaxing the molecules forming the SAMS and the top two metal layers until forces are below 0.04 eV/
Å. The exchange correlation SGGA.PBE functional has been used, with a (DZP) basis set for valence
cobalt electrons (i.e., to describe the valence molecular orbitals), a (2×8×1) k-point sampling and a
mesh cutoff of 100 Ha. The electronic structure and the SAM induced work function shift are then
computed following the same method as used for the calculations on the bare cobalt surface, with a
DZP basis set for the valence molecular orbitals. As described for the metal surface, a second ghost
layer of gold atoms has been also positioned at -0.1 Å below the top molecular atoms to reduce the
artificial push-back effect and prevent any artificial creation of bond dipole due to BSSE (Basis Set
�10
Page 51
Superposition Errors).29 These two effects may induce errors larger enough to prevent a direct
comparison between the work function shifts of the different systems.
The change in the work function upon SAM deposition, generally referred to as ‘vacuum level
shift’ is well studied in the literature30, 31 and has two main possible origins: (1) the intrinsic dipole
moment of the molecules, reduced by the induced dipole arising from the depolarization effects
triggered by the adjacent molecules;32 and (2) the interfacial electronic reorganization upon formation
of the Co-S bond resulting in a bond dipole in the vicinity of the interface. The apparent work function
shift is then simply described as the sum of a molecular contribution (ΔΦSAM) and a bond dipole
(ΔΦBD):
% (S4)
The relative intensities of these two contributions depend on whether we adopt a neutral or a radical
approach to describe the bonding of thiol to the metal surface. The neutral approach describes the
adsorption as the replacement of a bond (the S-H bond) with a S-Co bond whereas the radical scenario
considers the formation of a new covalent bond between the radical and the cobalt surface. For the sake
of consistency with our previous studies,33-35 we adopt here the ‘radical scenario’ for computing the
molecular contribution and the bond dipole. To this end, we compute the electrostatic potential profile
across a layer of radicals obtained by removing the cobalt atoms from the full system while keeping the
same geometry for the molecular part; we then compute the energy difference between the left and
right side of the layer, which corresponds directly to ΔΦSAM. The value of BD is obtained by
subtracting ΔΦSAM from the total shift. We note for sake of completeness that the amplitude of the
work function shift can be related to the corresponding dipole moment by the Helmholtz equation:36, 37
% (S5)
ΔΦ = ΔΦSAM +ΔΦBD
ΔΦ = µε0A
�11
Page 52
where µ is the dipole moment in the unit cell, projected along the axis normal to the surface, ε0 is the
vacuum permittivity and A is the area per dipole.
Co-AzBT molecular junctions. In a final stage, we add a second electrode on the top side of the
molecular layer to build a single heterogeneous molecular junction Co-AzBT/Au. This second
electrode is created by conserving the lattice parameters of the cobalt (111) surface and by converting
the cobalt atoms to gold atoms to avoid problems related to lattice incommensurability. The choice of
this type of electrode instead of a typical cobalt electrode is rationalized by two facts: (1) the
measurements were performed with a PtIr tip, which is not a ferromagnetic material; and (2) the work
function calculated for this top electrode is 4.81 eV, which is in good agreement with the experimental
PtIr work function of 4.86 eV.38 Note that this work function has been calculated by adding a ghost
layer of gold atoms at a distance of 1.7 Å from the top gold layer of the second electrode.
A van der Waals contact is assumed between the molecular layer and the top electrode, with an
interatomic distance determined as the sum of van der Waals radii of the hydrogen and gold atoms
(2.86 Å). The electronic transport calculations of the Co-AzBT/Au junctions were performed by the
combination of DFT to the Non-Equilibrium Green’s Function (NEGF) formalism,39 as implemented in
QuantumATK Q-2019.12-SP1 package.21, 22 This formalism has been widely recognized as a robust
tool to rationalize the experimental results and predict new features and trends for charge transport in
molecular junctions.35, 40, 41 The exchange-correlation potential is described with the SGGA.PBE
functional20 whereas the Brillouin zone was sampled with a (3×12×100) k-sampling. We expand the
valence electrons in a single zeta plus polarization basis set (SZP) for metal atoms (gold, gold ghost
and cobalt), and a double zeta polarization basis set (DZP) for the other atoms. The core electrons are
frozen and described by the norm-conserving Troullier−Martins pseudopotentials.23 As for SAMs
calculations, the platinum ghost atoms are described by a medium basis set and PseudoDojo potentials.
�12
Page 53
28 The mesh cutoff was set to 80 Ha with a temperature of 300 K. These parameters have been carefully
tested to ensure the convergence of the transmission spectrum.
The I-V characteristics have been calculated on the basis of the Landauer-Büttiker formalism,42
that links the transmission spectrum to the current in a coherent transport regime. When a bias is
applied, the current is calculated via the integration of the transmission spectrum within a bias window
defined by a Fermi-Dirac statistics in the left and right electrodes:
% (S6)
Where T(E) is the transmission spectrum, E the incident electron energy, f the Fermi-Dirac
function f(E-µ)=[1+exp(E-µ)/kT]-1, µL/R the chemical potential of the left/right electrode, e the
elementary charge, h the Planck constant, k the Boltzmann constant and V the applied bias, with
⎜µL - µR⎟ = eV.
It is important to note that for an accurate estimation of the current, the transmission
spectrum T(E) should be calculated in a self-consistent way for each bias. However, it is possible
to obtain a reasonable approximation for the current at low bias by using the transmission
spectrum at zero bias. This approximation is required here for our junctions with a large unit cell
and a spin-polarized electrode because the self-consistent calculations become very time
consuming. Accordingly, the current-voltage properties of Co-AZBT/Au were predicted by using
the transmission calculated at equilibrium.
I(V )= 2eh
T(E) f E − µL( )− f E − µR( )⎡⎣ ⎤⎦∫ dE
�13
Page 54
7. Junctions under bias
The computed ON/OFF ratios can be affected by the voltage drop profile across the junction.
This voltage drop is hard to guess and should be computed theoretically. However, as explained
in the methodology, these systems were computationally too heavy to be driven out of
equilibrium. To get a qualitative idea of the ratio expected from theory, we thus integrated the
equilibrium transmission by considering three extreme scenarios (see Fig. S5):
1. Under bias, one Fermi energy is fixed and the other one is driven up by e⎪V⎪.
2. Under bias, one Fermi energy is fixed and the other one is driven down by -e⎪V⎪.
3. Under bias, one Fermi level is driven up by e⎪V⎪/2 and the second level is driven
down by -e⎪V⎪/2.
Figure S5. Semilog scale plot of current (nA) versus bias when (1) a single Fermi level is driven
up by e⎪V⎪, (2) a single Fermi level is driven down by -e⎪V⎪ and (3) one Fermi level is driven up
by e⎪V⎪/2, and the second level is driven down by -e⎪V⎪/2.
With our simplified methodology, the first and second scenario should give rise to a significant
rectification as shown by the difference of the calculated I-V shapes between cases (1) and (2),
because the HOMO and LUMO levels are not located symmetrically with respect to the Fermi
energy (Fig. 6, main text), which is not seen experimentally. We thus conclude that the
�14
Page 55
symmetrical bias repartition, scenario (3), is more reasonable. Noteworthy, the higher slope that
appears beyond 0.3V for cis1 (see Fig S5-c) arises from the integration of the equilibrium
transmission spectrum, which is clearly an approximation. Since this feature is not seen in the
experimental curve, we speculate that the equilibrium transmission is not accurate beyond 0.3V
and hence that our theoretical ratios are more accurate below 0.3V. We report the predicted ratios
corresponding to the three scenarios in Table S3. In this low voltage range, the three scenarios
predict ratios between trans and cis1 between 14 and 53, hence with the same order of magnitude
as the experimental values.
Table S3. The calculated ON/OFF ratio between trans and cis1 forms for the different scenarios.
REFERENCES
1. Shirley, D. A., High-Resolution X-Ray Photoemission Spectrum of the Valence Bands of Gold.
Phys. Rev. B 1972, 5, 4709-4714.
2. Ulman, A., An introduction to ultrathin organic films : from Langmuir-Blodgett to Self-
assembly. Academic press: Boston, 1991.
Scenarios Ratios 0.1 V 0.2 V 0.3 V 0.4 V 0.5 V
(a)Trans1/cis1 17.8 17.3 18.9 20.6 22.4
Trans2/cis1 19.4 16.7 16.5 16.9 17.5
(b)Trans1/cis1 53.0 14.6 1.0 0.2 0.2
Trans2/cis1 26.8 5.9 0.4 0.2 0.2
(c)Trans1/cis1 34.7 44.1 33.8 14.7 4.2
Trans2/cis1 26.6 24.9 15.9 6.3 1.7
�15
Page 56
3. Parikh, A. N.; Allara, D. L.; Ben Azouz, I.; Rondelez, F., An intrinsic relationship between
molecular structure in self-assembled n-alkylsiloxane monolayers and deposition temperature. J. Phys.
Chem. 1994, 98, 7577-7590.
4. Engelkes, V. B.; Daniel Frisbie, C., Simultaneous nanoindentation and electron tunneling
through alkanethiol self-assembled monolayers. The Journal of Physical Chemistry B 2006, 110 (20),
10011-10020.
5. Smaali, K.; Desbief, S.; Foti, G.; Frederiksen, T.; Sanchez-Portal, D.; Arnau, A.; Nys, J. P.;
Leclere, P.; Vuillaume, D.; Clement, N., On the mechanical and electronic properties of thiolated gold
nanocrystals. Nanoscale 2015, 7 (5), 1809-1819.
6. Thomas, L. Etude par microscopie à sond locale des propriétés électriques de monocouches
auto-assemblées de molécules photo-commutables sur substrats ferromagnétiques. PhD thesis,
University of Lille, 2018.
7. Necas, D.; Klapetek, P., Gwyddion: an open-source software for SPM data anylsis. Central
European Journal of Physics 2012, 10, 181-188.
8. Horcas, I.; Fernández, R.; Gómez-Rodríguez, J. M.; Colchero, J.; Gómez-Herrero, J.; Baro,
A. M., WSXM: A software for scanning probe microscopy and a tool for nanotechnology. Rev. Sci.
Instrum. 2007, 78 (1), 013705.
9. Cui, X. D.; Primak, A.; Zarate, X.; Tomfohr, J.; Sankey, O. F.; Moore, A. L.; Moore, T. A.;
Gust, D.; Harris, G.; Lindsay, S. M., Reproductible measurement of single-molecule conductivity.
Science 2001, 294, 571-574.
10. Cui, X. D.; Zarate, X.; Tomfohr, J.; Sankey, O. F.; Primak, A.; Moore, A. L.; Moore, T. A.;
Gust, D.; Harris, G.; Lindsay, S. M., Making electrical contacts to molecular monolayers.
Nanotechnology 2002, 13, 5-14.
�16
Page 57
11. Morita, T.; Lindsay, S., Determination of Single Molecule Conductances of Alkanedithiols by
Conducting-Atomic Force Microscopy with Large Gold Nanoparticles. J Am Chem Soc 2007, 129 (23),
7262-7263.
12. Johnson, K. L., Contact Mechanics. Cambridge University Press: New York, 1987.
13. WebElements, https://webelements.com.
14. Thermo Scientific XPS: knowledge base. https://xpssimplified.com/periodictable.php.
15. NIST, X-ray Phtoelectron Spectroscopy Data Base, V. 3.5, https://srdata.nist.gov/xps/.
16. Moulder, J. F.; Chastain, J., Handbook of X-ray photoelectron spectroscopy: A reference book
of standard spectra for identification and interpretation of XPS data. Perkin-Elmer Corporation: 1992.
17. Hoertz, P. G.; Niskala, J. R.; Dai, P.; Black, H. T.; You, W., Comprehensive investigation of
self-assembled monolayer formation on ferromagnetic thin film surfaces. J. Am. Chem. Soc. 2008, 130
(30), 9763-9772.
18. Devillers, S.; Hennart, A.; Delhalle, J.; Mekhalif, Z., 1-Dodecanethiol Self-Assembled
Monolayers on Cobalt. Langmuir 2011, 27, 14849-14860.
19. Chuang, T. J.; Brundle, C. R.; Rice, D. W., Interpretation of the x-ray photoemission spectra of
cobalt oxides and cobalt oxide surfaces. Surface Science 1976, 59 (2), 413-429.
20. Perdew, J. P.; Burke, K.; Ernzerhof, M., Generalized Gradient Approximation Made Simple.
Phys. Rev. Lett. 1996, 77 (18), 3865-3868.
21. QuantumATK Q-2019.12-SP1, Synopsis QuantumATK, www.synopsys.com/silicon/
quantumatk.html.
22. Smidstrup, S.; Stradi, D.; Wellendorff, J.; Khomyakov, P. A.; Vej-Hansen, U. G.; Lee, M.-E.;
Ghosh, T.; Jónsson, E.; Jónsson, H.; Stokbro, K., First-principles Green's-function method for surface
calculations: A pseudopotential localized basis set approach. Physical Review B 2017, 96 (19), 195309.
�17
Page 58
23. Troullier, N.; Martins, J. L., Efficient pseudopotentials for plane-wave calculations. Physical
Review B 1991, 43 (3), 1993-2006.
24. Smidstrup, S.; Markussen, T.; Vancraeyveld, P.; Wellendorff, J.; Schneider, J.; Gunst, T.;
Verstichel, B.; Stradi, D.; Khomyakov, P. A.; Vej-Hansen, U. G.; Lee, M.-E.; Chill, S. T.;
Rasmussen, F.; Penazzi, G.; Corsetti, F.; Ojanperä, A.; Jensen, K.; Palsgaard, M. L. N.; Martinez,
U.; Blom, A.; Brandbyge, M.; Stokbro, K., QuantumATK: an integrated platform of electronic and
atomic-scale modelling tools. Journal of Physics: Condensed Matter 2019, 32 (1), 015901.
25. Neugebauer, J.; Scheffler, M., Adsorbate-substrate and adsorbate-adsorbate interactions of Na
and K adlayers on Al(111). Physical Review B 1992, 46 (24), 16067-16080.
26. Natan, A.; Kronik, L.; Shapira, Y., Computing surface dipoles and potentials of self-assembled
monolayers from first principles. Applied Surface Science 2006, 252 (21), 7608-7613.
27. Michaelson, H. B., The work function of the elements and its periodicity. J. Appl. Phys. 1977,
48 (11), 4729-4733.
28. van Setten, M. J.; Giantomassi, M.; Bousquet, E.; Verstraete, M. J.; Hamann, D. R.; Gonze,
X.; Rignanese, G. M., The PseudoDojo: Training and grading a 85 element optimized norm-conserving
pseudopotential table. Computer Physics Communications 2018, 226, 39-54.
29. Van Dyck, C.; Bergren, A. J., Large Built-In Fields Control the Electronic Properties of
Nanoscale Molecular Devices with Dipolar Structures. Adv. Electron. Mater. 2018, 4 (5), 1700656.
30. Heimel, G.; Romaner, L.; Zojer, E.; Bredas, J.-L., The Interface Energetics of Self-Assembled
Monolayers on Metals. Accounts of Chemical Research 2008, 41 (6), 721-729.
31. Heimel, G.; Rissner, F.; Zojer, E., Modeling the Electronic Properties of π-Conjugated Self-
Assembled Monolayers. Adv. Mater. 2010, 22 (23), 2494-2513.
32. Cornil, D.; Olivier, Y.; Geskin, V.; Cornil, J., Depolarization Effects in Self-Assembled
Monolayers: A Quantum-Chemical Insight. Adv. Funct. Mater. 2007, 17 (7), 1143-1148.
�18
Page 59
33. Crivillers, N.; Liscio, A.; Di Stasio, F.; Van Dyck, C.; Osella, S.; Cornil, D.; Mian, S.;
Lazzerini, G. M.; Fenwick, O.; Orgiu, E.; Reinders, F.; Braun, S.; Fahlman, M.; Mayor, M.; Cornil,
J.; Palermo, V.; Cacialli, F.; Samori, P., Photoinduced work function changes by isomerization of a
densely packed azobenzene-based SAM on Au: a joint experimental and theoretical study. Phys. Chem.
Chem. Phys. 2011, 13 (32), 14302-14310.
34. Osella, S.; Cornil, D.; Cornil, J., Work function modification of the (111) gold surface covered
by long alkanethiol-based self-assembled monolayers. Physical Chemistry Chemical Physics 2014, 16
(7), 2866-2873.
35. Rodriguez-Gonzalez, S.; Xie, Z.; Galangau, O.; Selvanathan, P.; Norel, L.; Van Dyck, C.;
Costuas, K.; Frisbie, C. D.; Rigaut, S.; Cornil, J., HOMO Level Pinning in Molecular Junctions: Joint
Theoretical and Experimental Evidence. J. Phys. Chem. Lett. 2018, 9 (9), 2394-2403.
36. Schuhmann, D., The model of two parallel condensers and the application of the Helmholtz
formula to adsorption of polar surfactants from aqueous solutions. Journal of Electroanalytical
Chemistry and Interfacial Electrochemistry 1986, 201 (2), 247-261.
37. Evans, S. D.; Urankar, E.; Ulman, A.; Ferris, N., Self-assembled monolayers of alkanethiols
containing a polar aromatic group: effects of the dipole position on molecular packing, orientation, and
surface wetting properties. J Am Chem Soc 1991, 113 (11), 4121-4131.
38. Lee, N. J.; Yoo, J. W.; Choi, Y. J.; Kang, C. J.; Jeon, D. Y.; Kim, D. C.; Seo, S.; Chung, H. J.,
The interlayer screening effect of graphene sheets investigated by Kelvin probe force microscopy. Appl.
Phys. Lett. 2009, 95 (22), 222107.
39. Brandbyge, M.; Mozos, J.-L.; Ordejón, P.; Taylor, J.; Stokbro, K., Density-functional method
for nonequilibrium electron transport. Phys. Rev. B 2002, 65 (16), 165401.
�19
Page 60
40. Schmaus, S.; Bagrets, A.; Nahas, Y.; Yamada, T. K.; Bork, A.; Bowen, M.; Beaurepaire, E.;
Evers, F.; Wulfhekel, W., Giant magnetoresistance through a single molecule. Nature Nanotech. 2011,
6, 185-189.
41. Van Dyck, C.; Geskin, V.; Kronemeijer, A. J.; De Leeuw, D. M.; Cornil, J., Impact of
derivatization on electron transmission through dithienylethene-based photoswitches in molecular
junctions. Phys. Chem. Chem. Phys. 2013, 15 (12), 4392-4404.
42. Büttiker, M.; Imry, Y.; Landauer, R.; Pinhas, S., Generalized many-channel conductance
formula with application to small rings. Physical Review B 1985, 31 (10), 6207-6215.
�20